Although remarkable advances have been made in the treatment of the acute leukemias, particularly resistant forms of leukemia remain. In 1999, 28,000 children and adults in the U.S. will be diagnosed with leukemia and 21,000 will die of their disease. This variability in clinical response is due in part to the tremendous heterogeneity of the disease itself. Traditionally classified solely on the basis of morphology and cytochemistry, the acute lymphoid or lymphoblastic leukemias (ALL) and the acute myeloid leukemias (AML) are characterized by highly variable clinical and biologic behavior, immunophenotypes, and chromosomal abnormalities. Striking differences in outcome may be seen in cases with the same cytogenetic profile, implying that more subtle genetic abnormalities also impact disease biology and response. We hypothesize that cDNA microarray technology will yield quantitative, orderly, and systematic gene expression profiles that can be used to design more clinically relevant classification schemes and to predict therapeutic response. By conducting correlative science studies accompanying NCI-sponsored clinical trials in children and adults affected by acute leukemia for the Pediatric Oncology Group, Children's Cancer Study Group, and Southwest Oncology Group, and by maintaining the largest leukemia tissue repositories in the world, we are poised to propose the following specific aims: 1. To Further Optimize cDNA Microarray Technology for Studies in Primary Human Leukemia Samples. 2. To Characterize the Molecular Variations Among Highly Selected Acute Leukemia Cases Using at Least 30,000 Genes. Cases have been selected using two approaches: 1) therapeutic response/resistance and 2) the presence of specific cytogenetic abnormalities. Study sets in AML include: 1) patients with "primary resistant" disease; 2) patients in long-term remission; 3) paired pre-treatment and relapse samples; 4) patients responding or failing specific treatment regimens; and 4) cases selected by genotype [t(8.21), inv(16), t(15;17), t(4;11), t(9;11), and complex]. In ALL, cases are being selected prospectively using two approaches: 1) the presence of residual disease vs. complete molecular response during the treatment course using automated quantitative molecular monitoring methods; and 2) by genotype [hyperdiploid, t(12;21), t(9;22), t(1;19), and t(4;11)]. 3. To Apply Multivariate Clustering Methods to Group Acute Leukemias That are Coherent in their Expression Patterns. 4. To Use Automated Quantitative "Real-Time" PCR Technologies to Validate cDNA Microarray Analyses. 5. To Use High Performance Computing and Informatics Technologies to Link Large Genomic Data Sets with Clinical Databases. All leukemia samples have associated clinical databases containing detailed patient information, laboratory data (cytogenetics, correlative scientific studies), and therapeutic response data. Our experienced clinical trials biostatisticians will work with the UNM High Performance Computing Center (a National Supercomputing Facility) and Sandia National Laboratory (both world leaders in massively scalable parallel computing, statistics, informatics, and visualization tools) to meet this aim.